Astrocyte and oligodendrocyte enhancer discovery from single cell epigenetics.

(A) Example astrocyte- and oligodendrocyte-specific peaks near the loci of astrocyte-specific gene AQP4 and oligodendrocyte-specific gene OPALIN, identified in human MTG snATAC-seq data33. (B) Differing approaches to identify candidate enhancers. Specific accessibility peaks are depicted as peaks, and specifically demethylated regions are depicted as troughs. Schemes not utilizing a particular data modality are shown as “Agnostic”. Marker gene selection criteria can use accessibility from either mouse or human. Icons represent identification schemes; gold star candidate enhancers undergo more stringent criteria than those with gold squares (see Methods for details). (C) Workflow for enhancer cloning, packaging, screening, and validation. Enhancers are cloned into a pAAV plasmid upstream of a minimal human beta-globin promoter and SYFP2 reporter, and plasmids are packaged into PHP.eB AAVs. Enhancer-AAVs are injected intravenously into retro-orbital sinus, and expression is assessed by imaging. Promising enhancer-AAVs then go on to secondary validation experiments consisting of cross-species validation, molecular characterization by IHC and/or multiplexed FISH, and flow cytometry for single cell RNA-seq. (D-E) Candidate identification schemes and summarized screening results for astrocyte-specific (D) and oligodendrocyte-specific enhancers (E). Overall 50% (25/50) of candidate astrocyte enhancers, and 49% (21/43) of candidate oligodendrocyte enhancers, show the intended specificity after intravenous delivery of PHP.eB AAVs. Testing result bar: Y = yes, enhancer-AAV gives strong or moderate on-target expression pattern; N = no, enhancer-AAV fails to express; W = weak on-target expression pattern; Mx = mixed expression pattern consisting of on-target cells plus unwanted neuronal populations; Off = off-target expression pattern; ND = no data. Note both enhancers giving strong/moderate (“Y”) and weak (“W”) specific expression are grouped here for overall success rate analysis.

Epigenetic characterization of candidate enhancers in additional chromatin accessibility datasets.

(A-D) Accessibility profiles of all tested candidate human astrocyte-specific (A, C) and human oligodendrocyte-specific (B, D) enhancers. Human enhancer regions are characterized in the datasets of Mich et al.33 (A-B), who performed snATAC-seq on neurosurgical MTG samples, and of Fullard et al.94 (C-D), who performed bulk ATAC-seq on neuronal (sorted NeuN+) and non-neuronal (sorted NeuN-) nuclei from dissections spanning multiple regions of human postmortem forebrain. Overall, many candidate astrocyte- and oligodendrocyte-specific enhancers show accessibility specific to non-neuronal cells across much of the human forebrain. For each genomic region we show their peak nomination scheme matching to Figure 1B, enhancer name. (E-F) Screening results from testing human candidate enhancers (repeated from Figure 1D-E, provided again for visualization). Testing result bar: Y = yes, enhancer-AAV gives strong or moderate on-target expression pattern; N = no, enhancer-AAV fails to express; W = weak on-target expression pattern; Mx = mixed specificities consisting of on-target cells plus unwanted neuronal populations; Off = off-target expression pattern, ND = no data. (G-J) Accessibility profiles for all tested candidate mouse astrocyte-specific (G, I) and oligodendrocyte-specific enhancers (H, J). Mouse enhancer regions are characterized in the VISp scATAC-seq dataset of Graybuck et al.34 (G-H) and in the full mouse cerebrum dataset of Li et al.56 (I-J). (K-L) Screening results from testing mouse candidate enhancers (repeated from Figure 1D-E, provided again for visualization).

Genomic coordinates, sequence characterization, and mouse screening results of all tested astrocyte and oligodendrocyte enhancers.

Calculations of parameters are as described in Methods section.

List of mice used for characterizing the activity of astrocyte and oligodendrocyte-specific enhancer-AAV reporter vectors.

“Mouse ID” is a unique six-digit identifier for each mouse. “Genotype” describes the genotype. “Enhancer ID” is a unique label for each enhancer or promoter element tested (grouped by colors). “Vector ID” is a unique label for the enhancer-AAV DNA packaged into PHP.eB for testing. “Vector Full Name” is a descriptive name of the elements in each enhancer-AAV. “Data Modality” refers to the data collected from each mouse: EPI epifluorescence, IHC immunohistochemistry, STPT serial two-photon tomography, scRNA-seq single cell RNA-seq (SMARTerV4). “Targeted Cell Population” is Astrocytes or Oligodendrocytes. “Fluor” is SYFP2 or iCre(R297T) used to recombine tdTomato in Ai14 mouse line. “Delivery Method” is retro-orbital (RO) or intracerebroventricular (ICV). “Publication” refers to this work (Mich et al.), or if the mice were also reported on in a previous work (Ben-Simon et al.35), or whether the enhancer or promoter was previously described and used here as a positive control (Lee et al.57, Gow et al.59, Jüttner et al.41).

A collection of astrocyte-specific enhancer-AAV vectors with varying regional specificities and expression densities.

(A-B) Astrocyte-specific enhancer-AAVs marking many astrocytes throughout most of the CNS. AiE0380h (A, n = 8 animals tested) and GFAP promoter (B, n = 3) mark many astrocytes throughout gray matter in FB, MB, HB, and CBX. (C) AiE0385m shows a regional pattern labeling astrocytes in cerebrum (CH) but not in MB, HB, or CBX (n = 6). (D-F) Astrocyte-specific enhancer-AAVs labeling astrocytes in lumbar SpC. AiE0380h (D) and GFAP promoter (E) label many astrocytes in SpC gray matter, but AiE0385m (F) does not label SpC astrocytes. (G-O) Positive confirmation of molecular astrocyte identity across brain regions. SYFP2+ astrocytes are colabeled with anti-Sox9 immunoreactivity in VISp, CBX, and Pons. Images in G-O represent n = 2 animals each assessed by IHC. (P) Quantification of specificity for astrocytes by astrocyte enhancer-AAVs. Specificity and completeness for astrocyte labeling by enhancer-AAVs was quantified by costaining with anti-Sox9 antibody in VISp, Pons, and CBX. Specificity is defined as the number of SYFP2+Sox9+ / total SYFP2+ cells x 100%. Completeness is defined as the number of SYFP2+Sox9+ / total Sox9+ cells x 100%. Brains from one to three mice per condition were analyzed, with range 131-827 cells counted (median 311) per brain region analyzed. AiE0375m-labeling was only quantified in the Purkinje cell layer of CBX, not in the granule or molecular layers. Specificity was also quantified by scRNA-seq, defined as the percentage of sorted SYFP2+ cells mapping as astrocytes within the VISp molecular taxonomy89. Overall, specificity is high for many astrocyte-specific vectors, with “Scattered” and “Weak” vectors showing low completeness, and “Regional” vectors showing more completeness in certain regions. One to three animals were analyzed per vector per modality, as indicated. The same IHC specificity data for GFAP promoter-SYFP2 are repeated in Figure 6B. (Q) Distinct astrocyte morphologies throughout the brain with AiE0387m enhancer-AAV targeting “Most of CNS”. Images were acquired on a serial two-phtoton tomography blockface imaging platform (STPT, TissueCyte, n = 4 tested). Abbreviations: CH cerebrum, dSTR dorsal striatum, CA1 cornu ammonis 1, CBX cerebellar cortex, SpC spinal cord, VISp primary visual cortex.

Full screening results of all candidate enhancer-AAVs targeting astrocytes.

We injected mice with the indicated enhancer-AAV vectors between P42 and P56, then after 3-4 weeks we harvested brains, sliced them on a sliding microtome with freezing stage at 30 µm thickness, co-stained the sections with DAPI, then mounted them with Vectashield Vybrance. Insets show a full cortical column from VISp (primary visual cortex), and in some cases also the labeling in MB (midbrain) or HB (hindbrain) or CBX (cerebellar cortex) is also shown. Astrocyte-specific enhancer-AAV vectors are broadly grouped by expression pattern into the following categories: “Most of CNS astrocytes”, “Regional” meaning present at medium-to-high levels in one or more broad brain regions but not all, “Scattered” meaning a few astrocytes are strongly labeled throughout the brain, “Weak” meaning many astrocytes throughout the brain are labeled at low level, “Mixed specificities” meaning one or more off-target neuron populations are also labeled in addition to astrocytes, and “No astrocyte expression” meaning failure to detect any clear astrocytes in these whole-brain sagittal images. These epifluorescent screening images represent n = 1 to 27 animals tested per enhancer (median 2 animals), and were sometimes taken on multiple different microscopes (see Table S2 for full summary of screened animals).

Distinct astrocyte-specific expression domains of AiE2120m and AiE2160m.

We injected mice with the indicated astrocyte-specific SYFP2-expressing enhancer-AAVs and performed whole-brain serial two-photon tomography using the TissueCyte platform88. These vectors display largely non-overlapping zones of astrocyte expression: AiE2120m is expressed in astrocytes within multiple forebrain structures including CTX, STR, OB, LSX, HPF, and TH, as well as MB, whereas AiE2160m is expressed in MB, CBX, and HB structures as well as complementary forebrain structures including HY, MSC, and GPe, and OB. In the OB AiE2120m is expressed in astrocytes within the granule cell layer, internal plexiform layer, and periglomerular cell layer, whereas AiE2160m is expressed in a complementary pattern of astrocytes within the external plexiform layer. Each image series shows one animal representative of two to three animals tested. Abbreviations: CTX cerebral cortex, STR striatum, OB olfactory bulb, LSX lateral septal complex, HPF hippocampal formation, TH thalamus, MB midbrain, CBX cerebellar cortex, HB hindbrain, HY hypothalamus, MSC medial septal complex, GPe globus pallidus, external layer.

A collection of oligodendrocyte-specific enhancer-AAV vectors with varying levels of expression.

(A-C) Oligodendrocyte enhancer-AAVs marking many oligodendrocytes throughout most of the CNS. AiE0410m (A, n = 7 animals tested), AiE0409h (B, n = 5), and AiE0400h (C, n = 4) label many oligodendrocytes throughout FB, MB, HB, and CBX, but at differing expression levels. (D-F) Oligodendrocyte enhancer-AAVs marking oligodendrocytes in lumbar SpC. AiE0410m (D), AiE0409h (E), and AiE0400h (F) mark oligodendrocytes in gray and white matter of SpC, but at different intensities. (G-O) Positive confirmation of molecular oligodendrocyte identity across brain regions. SYFP2+ oligodendrocytes are colabeled with CC1 immunoreactivity in VISp, CBX, and Pons. Two to three animals per enhancer-AAV were analyzed. (P) Quantification of specificity for oligodendrocytes by oligodendrocyte enhancer-AAVs. Specificity and completeness for oligodendrocyte labeling by enhancer-AAVs was quantified by costaining with CC1 antibody in VISp. Specificity is defined as the number of SYFP2+CC1+ / total SYFP2+ cells x 100%. Completeness is defined as the number of SYFP2+CC1+ / total CC1+ cells x 100%. Brains from one to three mice per condition were analyzed, with range 101-332 cells counted (median 147) per brain region analyzed. Specificity was also quantified by scRNA-seq, defined as the percentage of sorted SYFP2+ cells mapping as oligodendrocytes within the VISp molecular taxonomy89. One to three animals were analyzed per vector per modality, as indicated. Overall, specificity is high for many oligodendrocyte-specific vectors, with “Weak” vectors showing low completeness. (Q) Myelinating oligodendrocyte morphologies throughout the brain with AiE0410m. Sections were visualized with STPT (n = 2 tested). Abbreviations: SpC spinal cord, VISp primary visual cortex, CBX cerebellar cortex, LSX lateral septal complex, MY medulla.

Full screening results of all candidate enhancer-AAVs targeting oligodendrocytes.

We injected mice with the indicated enhancer-AAV vectors between P42 and P56, then after 3-4 weeks we harvested brains, sliced them on a sliding microtome with freezing stage at 30 µm thickness, co-stained the sections with DAPI, and mounted them with Vectashield Vybrance. Oligodendrocyte-specific enhancer-AAV vectors are broadly grouped by expression pattern into the following categories: “Strong oligodendrocytes”, “Weak” meaning many oligodendrocytes throughout the brain are labeled at low level, “Mixed specificities” meaning several off-target neuron or astrocyte populations are also present in addition to oligodendrocytes, and “No oligodendrocyte expression” meaning failure to detect any clear oligodendrocytes in these whole-brain sagittal images. These epifluorescent screening images represent n = 1 to 20 animals tested per enhancer (median 2 animals), and were sometimes taken on multiple different microscopes (see Table S2 for full summary of screened animals).

Transcriptomic identities of prospectively targeted astrocytes and oligodendrocytes.

(A-C) Groups of transcriptomically profiled single cells, as visualized by UMAP. Single cells labeled by various astrocyte- and oligodendrocyte-specific enhancer-AAVs (n = 1946 quality-filtered cells) were profiled from 47 brains in 47 independent experiments by SMARTerV489, one to three animals per enhancer. Libraries were aligned to mm10 and transformed into UMAP space for visualization, with coloring by enhancer (A), mapped taxonomic cell type cluster (B), and taxonomic mapping confidence (C). Overall CTX astrocytes group away from CTX oligodendrocytes as expected, and MB/HB astrocytes and Bergmann glia astrocytes group away from CTX astrocytes, consistent with recent results18. Note that AiE0381h- and AiE2160m-labeled astrocytes were dissected from MB/HB region, and AiE0375m-labeled Bergmann glia were dissected from CBX region, but the remainder of the cells were dissected from VISp. (D-E) Quantifications of taxonomic cell type cluster mapping by enhancer vector. Prospectively labeled astrocytes from all enhancer-AAV vectors dissected from VISp predominantly map to cluster “5112 Astro-TE NN_3”, whereas those from MB/HB dissections (AiE0381h and AiE2160m) predominantly map to cluster “5109 Astro-NT NN_2”, and AiE0375m-labeled astrocytes from CBX dissections predominantly map to cluster “5102 Bergmann NN”. In contrast, all prospectively labeled oligodendrocytes predominantly map to cluster “5158 MOL NN”. Cluster identities are from a recent whole mouse brain taxonomy study18. (F-H) De novo motif detection in astrocyte- and oligodendrocyte-specific enhancer sequences using MEME-CHIP61 identifies one strong consensus motif in each set of sequences (top). These de novo motifs were mapped against databases of known TF motifs using TomTom (bottom), which identified the top hits as the Zic family consensus motifs for astrocytes, and Sox family motif for oligodendrocytes (Sox10 shown). These TFs (Zic5 and Sox10) show highly specific expression differences between astrocytes and oligodendrocytes from prospective scRNA-seq profiling (H). *** non-parametric Wilcoxon rank-sum test W = 577624 (Zic5) or W = 9838 (Sox10), p < 1e-16 each. (I-N) Intrinsic SYFP2 expression from the indicated enhancer-AAVs after retro-orbital administration. Images were generated by STPT. Boxes in I and L correspond to K and N, respectively. Scale in I and K is 500 µm. (O-R) MERFISH data showing the distribution of three astrocyte cell types revealed by single cell gene expression from the whole mouse brain18. Abbreviations: CTX cerebral cortex, STR striatum, GPe globus pallidus, external segment.

Optimizing astrocyte and oligodendrocyte enhancer strength.

(A) Native Enhancer and for Enhancer_3xC2 vector designs. The central approximate third of the enhancer (the “Core2” element) is marked by dark hatches, and this element is triply concatemerized in the Enhancer_3xC2 vector. Alternatively, the first or third segment (“Core1” or “Core3”) may be concatemerized (determined empirically). (B-J) Dramatic increase in expression levels while maintaining specificity using Enhancer_3xC1/2 vector designs. Brains from mice injected with the Enhancer or Enhancer_3xC1/2 vectors were processed and imaged in parallel in these experiments. (H-J) Zoom in view of AiE0390m- and AiE0390m_3xC2-injected mouse VISp shows high specificity for morphological astrocytes throughout cortical layers in both cases. Data represent one to two animals per vector with parallel tissue processing and imaging. (K) Quantification of specificity for astrocytes by concatemer astrocyte enhancer-AAVs within VISp by IHC and scRNA-seq as described in Figure 2P. The same IHC data for AiE0390m_3xC2 are repeated in Figure 6B. (L-M) Direct correlated quantification of enhancer strength by flow cytometry and scRNA-seq, for both astrocyte-(L) and oligodendrocyte-specific (M) enhancer-AAVs. The left (blue) y-axis represents the log-transformed vector transgene reads per million in individual sorted scRNA-seq-profiled cells. The right (brown) y-axis represents the log-transformed SYFP2 signal intensity of positively gated vector-expressing cells observed on the flow cytometer, quantified as the fold signal of positive cells normalized to non-expressing cell autofluorescence (taken as background). Points represent individual cells observed by scRNA-seq and by flow cytometry, visualized also as violins, and with the horizontal bar representing mean expression levels across all cells expressing that enhancer-AAV, across one to three replicate experiments per vector. Across all experiments, we observe significant correlation between mean expression intensity at the RNA level by scRNA-seq, and mean SYFP2 reporter expression by signal intensity (astrocytes: n = 26 experiments, Pearson correlation coefficient [PCC] 0.63, t = 3.97, df = 24, p = 0.00057; oligodendrocyte n = 22 experiments, PCC 0.53, t = 2.82, df = 20, p = 0.011; correlation t-tests). Furthermore, 3xC astrocyte enhancers are among the strongest enhancers we have characterized, typically several fold stronger than their native counterparts. One to three animals were analyzed per enhancer-AAV vector (the same dataset described in Figure 4A). Abbreviations: VISp primary visual cortex.

Sorting enhancer-AAV-labeled astrocytes and oligodendrocytes.

Example gating strategies for sorting AiE0390h_3xC2-labeled astrocytes and AiE0396h-labeled oligodendrocytes from mouse VISp. Each flow gating strategy represents one animal tested for that vector (n = 26 total animals for astrocytes and n = 21 for oligodendrocytes).

Robust specificities of astrocyte enhancer-AAVs with a difficult transgene cargo.

(A-B) Specific delivery of a difficult transgene cargo with an optimized astrocyte-specific transgene. Mice received the indicated enhancer-AAVs driving either SYFP2 or NeuroD1-mCherry by intravenous PHP.eB AAVs (5e11 gc per animal), and expression analyzed in brain by IHC with NeuN neuronal marker or Sox9 astrocyte marker. Specificity for neurons and astrocytes was quantified as (Reporter+[NeuN or Sox9]+ cells) / (All Reporter+ cells) in primary visual cortex (VISp, Layer 5 cells shown, n=2-3 animals per condition). The same IHC specificity data for GFAP promoter-SYFP2 and AiE0390m_3xC2-SYFP2 are repeated in Figures 2P and 5K, respectively. Two to three animals were analyzed per condition (each dot represents one animal). *** ANOVA with Tukey’s post hoc adjustment for multiple comparisons, F = 647, p < 1e-7, for GFAP promoter-driven NeuroD1-mCherry compared to every other condition.

Predictability of astrocyte enhancer-AAV expression patterns across body organs.

(A) Accessibility profiles of astrocyte-specific enhancers in the human whole-body accessibility atlas70. Single-cell profiles were grouped within each tissue into pseudo-bulk aggregates, then normalized according to the signal (reads in peaks) within the dataset. Accessibility profiles are likely to predict enhancer activities within each tissue. Focusing on liver, some astrocyte-specific enhancers are predicted to have “Strong” expression, and some are predicted to have very “Little or none” expression. In contrast, accessibility atlases do not predict expression of GFAP promoter across tissues. (B) Whole livers from mice injected intravenously with AiE0371h- and AiE0387m-enhancer-AAV vectors, stained with anti-GFP antibody. AiE0371h has high liver accessibility, is predicted to have high liver expression, and shows many strong SYFP2-expressing hepatocytes throughout the liver as predicted. In contrast, AiE0387m has little liver accessibility and so is predicted to have little liver expression, and in fact shows few positive SYFP2-expressing hepatocytes as predicted. (C) Agreement between liver expression predictions and liver expression measurements across several astrocyte-specific enhancer-AAV vectors. AiE0371m, AiE0371h, AiE0381h, and AiE0386m all show many SYFP2-expressing hepatocytes as predicted. AiE0390m, AiE0390h, AiE0375m, AiE0387m, and AiE0380h all show few weak SYFP2-expressing hepatocytes as predicted. GFAP promoter shows few but strongly expressing hepatocytes, which was not predictable from the accessibility atlases. Liver images in B and C represent two to four mice analyzed for each vector (each dot represents one mouse). *** non-parametric Wilcoxon rank sum test, W = 153, p < 1e-4 for the comparison of Strong versus “Little or none”-predicted enhancers.

Predictability of astrocyte enhancer-AAV expression patterns across disease states.

(A-C) Testing fidelity of enhancer-AAV expression across disease states. We used Dravet syndrome model Scn1a+/- mice to induce epilepsy-associated hippocampal gliosis, injected enhancer-AAVs prior to the critical period, and analyzed tissue for expression patterns after the critical period (A). We assessed hippocampal gliosis with anti-GFAP antibody and enhancer-AAV expression with anti-GFP antibody (B). AiE0390m maintained specific expression and similar levels in hippocampal astrocytes regardless of epileptic gliosis. In contrast, GFAP promoter expression strongly increased in gliotic astrocytes, and also was observed in dentate gyrus granule cells. Red dashed rectangles indicate the position of the expanded zoomed view, and the curved red arrows indicate a rotated view for one zoom. (C) Quantification of DGC labeling reveals a much greater number of DGC neurons labeled with GFAP promoter in Dravet syndrome model mice. Two to four animals were analyzed per condition (each dot represents one mouse). ** ANOVA with Tukey’s post hoc adjustments, F = 16, p < 0.01 for the GFAP promoter in Scn1a+/-mice compared to each other condition. Abbreviations: ML molecular layer, GCL granule cell layer, PoL polymorphic layer.

Astrocyte specific sensing of cholinergic signals in the nucleus accumbens during behavior.

(A) AiE0390m_3xC2 driving expression of iAChSnFR. Enhancer vector is cloned into a pAAV plasmid and packaged into PHP.eB AAVs. (B) Coronal section showing stereotaxic injection of enhancer virus expressing iAChSnFR in the nucleus accumbens (injection coordinates: AP: 1.2, ML: 1.3, DV: 4.1). (C) Behavior and imaging experiment setup. Top: dynamic foraging behavior task schematic. Bottom: Fiber photometry instrumentation schematic and fiber location in a coronal section. (D) Fiber photometry signals of acetylcholine fluctuations during task performance. Top: ∼30 min segment of a 2-hour session of dynamic foraging. Black dots represent the auditory cue, red dots represent time of first lick, blue dots represent water reward delivery. Bottom: 100 second (980-1080 seconds) zoom in on above session with 6 individual trials (4 rewarded and 2 unrewarded trials). Data from one mouse shown, representative of three mice assayed. (E) Trial-averaged signals of rewarded and unrewarded trials aligned to time of first lick (mean±sem).

Astrocyte- and oligodendrocyte-specific Cre AAVs.

(A) Design and testing of astrocyte- and oligodendrocyte-specific attenuated Cre-expressing enhancer-AAVs. (B-D) Specific Cre recombination in astrocytes driven by AiE0390m_3xC2. Ai14 reporter recombination is observed in astrocytes throughout the mouse brain (B). Recombination within VISp is highly specific (C), as evidenced by highly specific and complete recombination in Sox9+ astrocytes (D). (E-G) Specific Cre recombination in astrocytes driven by AiE0410m. This vector also includes three 3’UTR mir126 binding sites to eliminate unwanted expression in mir126-expressing endothelial cells75. Ai14 reporter recombination is observed in oligodendrocytes throughout the mouse brain (E). Recombination within VISp is highly specific (F), as evidenced by highly specific and complete recombination in CC1+ oligodendrocytes (G). (H-I) Quantification of recombination specificity (H) and completeness (I). Specificity is defined as (tdTomato+Sox9 or CC1+ cells) / (all tdTomato+ cells) x 100%. Completeness is defined as (tdTomato+Sox9 or CC1+ cells) / (all Sox9 or CC1+ cells) x 100%. Two to three animals were analyzed for each condition (each dot represents one mouse).

Genetic targeting of astrocytes and oligodendrocytes across species.

(A) Testing enhancer-AAV vectors by neonatal rat ICV injections. (B-C) Validation of astrocyte-specific enhancer-AAV vectors in rat by IHC with Sox9 antibody. AiE0390m_3xC2 vector also incorporates 4X2C 3’UTR miRNA binding sites to prevent any off-target labeling in excitatory neurons76. Each image represents one animal tested. (D) Validation of oligodendrocyte-specific enhancer-AAV vector in rat. AiE0641m shows specific expression in CC1+ VISp oligodendrocytes. Each image represents one animal tested. (E) Multiple stereotactic intraparenchymal injections into macaque brain. (F-I) Prospective labeling of macaque oligodendrocytes in vivo. AiE0410m enhancer-AAV vector gives widespread labeling of oligodendrocytes throughout the depth of motor cortex (F). Most labeled macaque oligodendrocytes exhibit multipolar ramified morphology indicative of local axon myelination (G). Some labeled oligodendrocytes exhibit morphologies suggesting wrapping around wider tubular structures highlighted with dashed white lines (H). SYFP2-expressing cells of both morphological types express the oligodendrocyte/OPC marker SOX10 with high specificity (I). One animal tested. (J-Q) Prospective labeling of macaque astrocytes in vivo. AiE0390m enhancer-AAV vector gives widespread labeling of astrocytes throughout the depth of somatosensory cortex (J). A few large Layer 5 pyramidal neurons are also labeled. Labeled astrocytes show the expected bushy morphology and GFAP immunoreactivity of astrocytes in parenchyma (K) and sometimes reside near walls of large-diameter tubular structures (L). SYFP2-expressing astrocytes express the astrocyte marker FGFR3 with high specificity (M). In another animal, an injection of AiE0390m_3xC2 vector with 4X2C 3’UTR miRNA binding sites gives high specificity (99%, N) for SOX9+ astrocytes in MO (O), without any large Layer 5 pyramidal neurons labeled. In another animal, AiE0390h vector gives high specificity (98%) of labeling for SOX9+ astrocytes (P-Q). Each vector represents one animal tested. Abbreviations: VISp primary visual cortex, MO motor cortex, SS somatosensory cortex.

Diverse morphologies of macaque astrocytes labeled by enhancer-AAVs.

(A-C) Labeling of both gray matter protoplasmic astrocytes and white matter fibrous astrocytes by AiE0390m enhancer-AAV. We show full cortical column of a somatosensory cortex injection site in A, with expanded insets to show protoplasmic astrocytes in gray matter (B) and fibrous astrocytes in white matter (C). Images represent one animal tested. (D) Confirmation of astrocyte identity by mFISH. Fibrous astrocytes in white matter express the astrocyte marker FGFR3, similar to gray matter protoplasmic astrocytes (Figure 8M).